摘要
以合金元素名称、合金元素含量和制备方法为输入层参数,以形状记忆性能为输出层参数,构建3×12×1三层BP神经网络模型,同时对预测能力和预测精度进行了验证,并对模型优选出的四元钛镍合金进行了显微组织、物相组成、形状记忆性能和耐腐蚀性的测试。结果表明,该BP神经网络模型的预测能力强、预测精度高,最大预测误差2.3%,优选出的四元钛镍合金为自蔓延合成Ti-50Ni-0.2Co-0.1Y合金,形变回复率高达99.9%,腐蚀电位较常规熔炼法制备Ti-50Ni钛镍合金正移442m V。
BP neural network model with three layers of 3×12×1 was built with the name of alloying elements, content of alloying elements and method of preparation as input parameters, and with shape memory property as output parameter 三.The checkout test was given out. Moreover, the microstructure, phase composition, shape memory property and corrosion resistance of the optimized Ti Ni alloy were tested. The results show that BP neural network model has good prediction ability,high accuracy of prediction, and with the maximum prediction error of 2.3%. Furthermore, the optimized quaternary alloy is Ti-50Ni-0.2Co-0.1Y alloy by self-propagating synthesis,and the deformation recovery rate of the optimized alloy could reach by 99.9%, moreover, compared with Ti-50 Ni titanium nickel alloy prepared by conventional smelting method. The corrosion potential of the optimized alloy prepared by self-propagating synthesis changes to the positive shift by 442 mV.
出处
《热加工工艺》
CSCD
北大核心
2015年第14期76-79,共4页
Hot Working Technology
基金
2014年度广西科学研究与技术开发计划项目(桂科攻14122007-20)
2014年柳州市科学研究与技术开发计划项目(2014J040402)
2014年度柳州市科学研究与技术开发计划项目(2014C010203)